palaeontologists to escape from the risk of circularity in using internal measures of the
fossil record to assess the quality of that same fossil record?
Clade versus age techniquesIndependenceA sideways leap provides a partial answer, and that is to use a source of data on the history
of life that is independent of the rocks, namely phylogenetics. It has been argued (Platnick
1979; Patterson 1982; Smith and Littlewood 1994; Benton and Hitchin 1996, 1997) that
phylogenetic data, whether from the cladistic analysis of morphological characters or from
molecular phylogenetic reconstruction, are essentially independent of stratigraphic
(geological age, rock distribution) evidence.
In cladistics, characters are determined and polarized (primitive•derived) according to
their distribution among a group of organisms, living and extinct, and without reference
to geological age. Many analysts these days do not even polarize the characters, so they
include no directional information prior to the analysis. Molecular phylogeny
reconstruction is even more obviously divorced from stratigraphy in that all the organisms
under investigation are extant, and characters are generally not polarized. The implied
history of the group is then subtended from the present-day with no reference to fossil
taxa.
Trees, whether cladistic or molecular, cannot entirely escape from time-related
aspects. A small input of stratigraphic bias may be associated with the choice of outgroups
(comparator standards), but outgroups can readily be changed and analyses re-run. In
addition, there are certain unavoidable temporal biases in phylogenies that relate to their
geometry and the completeness of taxon sampling (Wagner 2000a) and to the relative
timings of acquisition of apomorphies and homoplasies (Wagner 2000b). These issues do
not substantially modify the geometry of a tree and the relative order of branching points,
and hence independence between stratigraphy and tree shape is sustained (Benton 2001;
Wills 2002). More serious though is that the age versus clade measures cannot detect
major hiatuses in the rock record. So, if all early members of a clade are unknown as
fossils, the relative order of appearance of lineages within that clade then becomes
meaningless.
The hypothesis behind all age versus clade comparisons is that congruence indicates the
true pattern. It is accepted of course that the fossil record is subject to bias, as are the
techniques of cladistics and molecular phylogenetics. But the biases that might affect these
three approaches are clearly different, and unlikely to reinforce each other. So, if a
phylogenetic tree is congruent with the order of fossils in the rocks, it is most likely that
both are correct. If incongruence is found, then it cannot be said whether the tree or the
fossils, or both, are incorrect. The test for congruence does not mean that the entire tree
is congruent with stratigraphy, but that it is more congruent than random, and certainly
not significantly incongruent.
82 THE QUALITY OF THE FOSSIL RECORD